41257 Research on the statistics of grain lattice echoes and their use in grain size estimation and grain echo suppression : Karpur, P.; Newhouse, V.L.; Disque, R.C. Drexel University, Philadelphia, Pennsylvania, (United States), AD-A199 811/1/GAR, 170 pp. (19 Aug. 1988)

Abstract

Abstract : Split spectrum processing (SSP) was developed in the late seventies for ultrasonic applications as a convenient method of introducing frequency diversity in ultrasonic signals for the purpose of the reduction of coherent material noise. Although the technique is very powerful, the use of SSP remained somewhat limited because of the ambiguity that prevailed in the selection of suitable processing parameters. The ambiguity was mainly because of a lack of a physical model of the spectral splitting process. The need for such a physical modeling has been fulfilled as a part of this dissertation. An ultrasonic signal has been modeled as a time limited process which facilitates the use of the frequency sampling theorem to describe the frequency splitting process. Further, the theory of Fourier integrals is used to determine the bandwidth of the theoretical SINC filters obtained as per the frequency sampling theorem. Experimental results have been provided to corroborate the theory. Another aspect of SSP addressed by this thesis is a detailed theoretical analysis of a newly developed algorithm called the Polarity Thresholding (PT) algorithm. Yet another important aspect of SSP addressed by this dissertation is the demonstration of the utility of SSP in many engineering materials such as centrifugally cast stainless steel (CCSS), composites and interface between bimetallic alloys. The potential of SSP as a tool for material characterization is also experimentally demonstrated. Theses.